Table I.
a
Results for Analysis of Mixtures of Lead(ll), Cadmium(ll), and Tin(lV) Millif aradays Added Founda Constituent Recovery, 7' 0.9775 0.9771 Pb(I1) 99.96 1.0210 1 .0220 100.1 Cd(I1) 0.8161 0.8153 99.90 Sn(1V) Average of eight values.
Mixtures of lead(II), cadmium(II), and tin(1V) were analyzed as follows, and the results are given in Table I. Lead(I1) and cadmium(I1) were removed by reduction from ammoniacal solution using a mercury cathode a t -0.90 volt. The amalgam was removed through a stopcock located a t the bottom of the flask. Beads of amalgam remaining in the flask were removed by gathering with three successive 5-ml. portions of clean mercury and draining through the stopcock being careful not to entrain any aqueous phase. Sufficient clean mercury was added to form a new working electrode, and concentrated HBr was added to neutralize the base and provide a 0.2M excess of HBr. Enough solid NaBr was added so that its concentration would be about 3 X 1 and tin(I1) was reduced and oxidized as previously described. The lead cadmium amalgam was placed in a clean titration cell along with sufficient 3M NaBr 0.25M tartaric acid
+
+
solution, and the system was reduced a t -0.85 volt until a constant low background current was obtained. Then the analysis was performed using the procedure developed for mixtures of these two elements. The method was applied to the determination of tin in a National Bureau of Standards brass sample, NBS-37d1 that contained about equal quantities of lead and tin. The sample was dissolved in a mixture of nitric acid, sulfuric acid, and sodium bisulfate, and fumed to remove the nitrate. After cooling, the sample was dissolved in 0.5M am0.35M ammonia; monium tartrate and the pH was adjusted to 9.0 with sodium hydroxide. Copper(I1) was removed at -0.60 volt. The mercury was removed, and a fresh portion was added. The solution was electrolyzed a t -1.35 volt to remove lead(I1) and nickel(I1). In the presence of fresh mercury and excess HBr and NaBr, tin(1V) was reduced to tin(I1). The tin
+
was determined as mentioned before. Four values were obtained ranging from 0.96 to 0.98% tin with an average of 0.97y0. These values are in closer agreement with the certificate value (0.97%) than when the tin was precipitated as metastannic acid, dis0.2M HBr, and solved in 3M NaBr determined by potentiostatic coulometry (11). Probably some lead was interfering with the latter method.
+
LITERATURE CITED
(1) Alfonsi, B., Anal. Chim. Acta 22, 431 (1960). (2) Ibid.. 23. 375 (1960). (3j Ibid.; 25; 274 (i96ij. (4) Ibid., p. 374. (5) Ibid., 26, 316 (1962). (6) Baumgarten, S., et al., Ibid., 20, 397 (1959). (7) Lingane, J. J., IND. ENG. CHEM., ANAL.ED. 15, 583 (1943). (8) Lingane, J. J., ANAL. CHEM.18, 429 (1946). (9) Ibid., 23, 1798 (1951). (10) Page, J. A., Maxwell, J. A., Graham, R. P., Analyst 87, 245 (1962). (11) Wise, W. M., Williams, J. P., ANAL. CHEM.37, 1292 (1965). W. M. WISE
D. E. CAMPBELL Research and Development Laboratories Corning Glass Works Corning, N. Y. Presented at the Seventeenth Annual Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, February 1966.
Nondestructive Activation Analysis of Crude Oils for Arsenic to One Part per Billion, and Simultaneous Determination of Five Other Trace Elements SIR: A sensitive and reliable method of analysis for 1 to 100 p.p.b. of arsenic in crude oils is needed in petroleum technology because of the possible effect of arsenic on expensive platinum reforming catalysts. The decrease in hydrogenation activity of platinum metal in aqueous systems in the presence of arsenic (13, 17), and the deposition of arsenic on the platinum catalyst during the reforming process (18) may indicate that in certain forms arsenic has a specific poisoning effect in hydrocarbon media similar to that observed in aqueous systems but opinion on this point is divided (5, 9, 10, 11). The amount of arsenic recovered from the used catalyst indicates that the concentration of arsenic in platformer feedstocks is generally less than 10 p.p.b. All of the existing wet chemical methods for the analysis of arsenic in hydrogen media a t less than 100 p p b . depend upon the destruction of the hydrocarbon and conversion of the 1080
ANALYTICAL CHEMISTRY
arsenic to ionic form (1, 12, 16). This process is undesirable because arsenic may be lost at the elevated temperatures generally used for ashing the sample (15), because a relatively high arsenic blank results from the reagents in ashing the volume of oils required for suitable sensitivity, and, finally, because the ashing of petroleum under conditions to reduce loss of trace elements is tedious and time-consuming. Smales (19) has shown that thermal neutron capture by As75 to produce (26.4 hr.; 0.56-and 1.21-m.e.v. 1 ; 0.4and 2.97-m.e.v. 0-) is very sensitive for arsenic. This reaction has been used to determine arsenic in a wide variety of matrices (2, 4, 6, 8 ) including reforming catalysts (18). All of these activation analysis methods are destructive of the sample and rely upon chemical separation of the arsenic activity which is generally determined by beta counting. Such a procedure is only reliable when the carrier is in the same chemical form as the element being analyzed. Re-
cently Girardi and Pauly (7) described a nondestructive method for higher levels of As than the one to be described. In our method arsenic is determined without destruction of the sample and gamma-ray spectrometry is used to measure the 0.56-m.e.v. gamma ray from As76 selectively and without chemical separation from the Cu, Zn, Na, Nil and Br activities which are determined simultaneously. EXPERIMENTAL
Counting Apparatus. Two axially mounted, horizontally opposed 3-inch by 3-inch sodium iodide scintillation crystals with a 7/8-inch space between for beta shields and the introduction of samples were in turn shielded by 8 inches of steel for gamma-ray counting. The balanced outputs from these two crystals were fed simultaneously to a RIDL 400-channel analyzer for gammaray spectrometry. The silver foils used to monitor the neutron flux were counted separately with a Geiger tube having a 179 mg./sq. cm. aluminum absorber between it and the foils.
Calibration. The energy scale of the R I D L analyzer was calibrated with a standard T1208 source (2.615 and 0.583 m.e.v.) located between the two detector crystals so that the gamma-ray peaks always fell in the same channels and the outputs from the two crystals were balanced so that the resolution of the higher energy peak was never greater than 5.5%. Procedure. Irradiations were performed in reactors located a t University of Buffalo and McMasters University in Toronto on 20 samples at a time for 10 minutes at a thermal flux of 10i3n/sq. cm. second or for 1 hour a t 101*n/sq. cm. second. Sample containers were 7-ml. snap-cap polyethylene vials or 5-ml. quartz ampoules with a 25% head-space above all samples. The polyethylene vials were used only for aqueous and for semisolid petroleum samples; the tops were heat sealed and each had a silver foil monitor sealed into the bottom. All other samples were weighed into quartz ampoules, frozen, evacuated, sealed under vacuum, and inserted into a 7-ml. polyethylene vial with a weighed silver monitor foil in the bottom. After irradiation the samples were returned to the local laboratory where the monitor foils were removed and counted, the samples in polyethylene were counted without further preparation, and the quartz ampoules were frozen, opened, warmed to room temperature, and quantitatively transferred to inactive vials before being counted. Samples were counted for 10 minutes each; the spectra were examined visually and recorded digitally. The integrated area of the As76 0.56 m.e.v. gamma-ray peak above the continuum was calculated from the digital data readout in a similar manner to that described by Soltys and Morrison (20) and corrected for analyzer deadtime, for the decay since irradiation, and normalized to the corrected silver monitor foil counts. The arsenic concentration in the samples was determined by comparison with the specific activity of pure arsenic solution standards. RESULTS AND DISCUSSION
Sensitivity. The use of a remote reactor imposes certain limits on the sensitivity of this method. For example, the size of the samples must be kept small for economy of shipping and irradiation costs. We found t h a t sensitivity could not be increased by longer irradiation times than those given because the hydrogen pressure from the radiolysis of hydrocarbons and possibly gamma-ray heating is sufficient to rupture the quartz ampoules. Radiation damage is a function of hydrocarbon structure and is greater for aliphatics than for high boiling aromatics and heterocyclics. For example, when ndecane samples were irradiated for 1 hour a t 10-1~n/sq. cm. second while suspended directly in the moderating water of the reactor, half of the ampoules ruptured before being re-
Table 1.
Blank Determinations of Arsenic in p.p.b. %-Decane99.49 Mol %
Set Sample Successive (1) determinations ( 2 ) (3) (4) (5) Av. Overall av. 0.5 Std. dev. f 0 . 4
0.5
0.2
0.8
0.3
11. Natural 182 181 5.1 4.1 5.2 3.4 4.0 6.0 4.2 5.8 5.7 3.9 0.5 0.5 5.2 3.4 $0.1 -0.7
Table Sample No. Arsenic added. D.D.b. Arsenicfound,’p.p.b.(l) (2)
(3) Av. Blank Netarsenicfound,p.p.b. Difference, p.p.b.
moved from the reactor, while the other half exploded violently when opened even though frozen and a t liquid temperature. The type of samples irradiated in polyethylene was less subject to radiolysis and the smaller quantity of hydrogen produced diffuses through the walls of the vial rather than accumulating and causing rupture. The 10 to 33 hours required to return the samples from the reactor further limits the sensitivity and when multiple samples are simultaneously irradiated there is an even greater decay on the last samples counted. By limiting the number of samples per irradiation to 20, we were able to complete duplicate 10minute counts on each one within two consecutive 8-hour shifts. The sensitivity or limit of detectability of the method is indicated in Table I by a series of blank determinations made of 4 samples of Research Grade ndecane from 3 different irradiations performed over a period of one month. The average arsenic content was only 0.5 p.p.b. with a standard deviation of =t0.4. No single value out of 11 differed from the average by more than 0.5 p.p.b. although the range in values is 1 p.p.b. On the basis of these
Copper Arsenic Bromine Nickel zinc Sodium
B
A 1 0.1
2
1.o
C 4 0.7 0.9
3 0.1 0.1 0.1 0.2 0.4 0.2
0.8
Arsenic in n-Decane 183 184 185 186 187 9.6 4.8 4.7 5.2 5.2 10.2 4.6 5.6 5.1 5.4 9.0 4.5 4.6 5.8 5.8 10.4 5.1 4.3 5.6 5.3 9.9 4.7 4.8 5.5 5.5 0.5 0.5 0.5 0.5 0.5 9.4 4.2 4.3 5.0 5.0 - 0 . 2 -0.6 -0.4 -0.2 -0.2
188 5.2 5.2 6.2 5.6 5.7 0.5
5.2 0.0
results the sensitivity is a t least 1 p.pb. and a single determination at this level will be within 1 p.p.b. of the true value 90% of the time. Accuracy. T o determine the accuracy of the method for naturally occurring arsenic a t the level of interest, a known arsenic-rich crude oil, (analysis by chemical and activation techniques = 770 f 10 p.p.b.) was diluted about 150-fold with the blank n-decane referred to above to form solutions containing less than 10 p.p.b. of arsenic with about the same boiling range as platformer feedstocks. These dilutions were made individually directly in the ampoules used for irradiation to reduce the reported “plab ing-out” or adsorption of arsenic at very low concentrations by the walls of the containing vessel. These dilute samples known to contain 4 to 10 p.pb. of naturally occurring arsenic were then analyzed by the procedure described and the results are summarized in Table 11. At this level no results differed from the expected value by more than 1 p.p.b. On the bases of these results arsenic can be determined a t the 4 to 10 p.p.b. level with an accuracy of 1p.p.b.
Table 111. Standards Reproducibility Concentration, rn absence Detectability p.p.m. interferences limit, p.p.b. 162 f6% 0.5 150 to 1200 f9% 0.1 9.8 f4% 10 352 f7Vn 2 40 f15% 1 16.76 0.3 f 10%
?Ray peak m.e.v.
VOL 38, NO. 8, JULY 1966
0.51 0.56
0.77 n si -.--
1.11
2.75
1081
Overall Reproducibility of Method. Eleven samples of a standard aqueous arsenic solution of the same size as the oil samples were analyzed as described, with a similar, more concentrated solution of arsenic as a reference. The results show that, despite a 50% range in the flux as indicated by the silver monitor foils,
the standard deviation of the arsenic found in the 100 p.p.b. solution was less than *8 p.p.b. Interfering Elements. The most serious possible interferences arising from thermal neutron capture are Sb122, 77.2 hr., 0.56 m.e.v. y ; BrE2, 35.9 hr., 0.55 m.e.v. y : and C U ~ 12.8 ~ , hr., p. Bromine can also interfere directly through the reaction, Br79(n, a) As76. The reaction has been shown to occur with samples of bromine that contain no arsenic, but the extent has not been measured and will depend on the fast neutron flux of the reactor or neutron source used. The contribution to the arsenic from this source was within experimental error as will be shown in the discussion of Table IV. The decay of the 0.56-m.e.v. gamma-ray peak was followed on all crude oils analyzed but there was no deviation from the half life of As76 to indicate the presence of antimony. The similarity in the sensitivity of the method for arsenic and antimony together with the similarity in their chemical behavior toward platinum catalysts may make their separate determination unnecessary. Both copper and bromine occasionally occur in petroleum fractions in great enough concentration to interfere. Other elements yield radionuclides with halflives and energies that may inter-
Table IV. Correction for Bromine and Copper Interferences
As Added Found, av. 6 values Added Found, av. 6 values Added Found, av. 5 values
Added Found, av. 2 values Added Found
p.p.b.
Br
Uncorrected values As, p.p.b.
9.2
26
10.7 9.0
27 132
15
8.9 9.4
130 911
26
9.3
1000
116
As
Cu
3.6
994
8.9 3.8 3.5
1061 1607 1683
10.5 5.4
Table V.
Method Activation Emission Activation Emission Activation Emission Activation Emission Activation Emission Activation Emission Activation Emission Activation Emission
Saniple Mid East-1 Mid East-2 Texas Venezuela Mississippi Kansas-1 Kansas-2 California
Table VI.
cu C
